cs.LGNov 10, 2026

Auditing the Judge: Internal Signals for LLM Reasoning Assurance

Authors: Mayank Gulati, Gerhard Wunder

Organizations: Freie Universität Berlin

Abstract

Language-model systems use self-judgment to decide which answers to release, review, or recompute. Because intervention is costly, an assurance score must rank errors well at the available budget, not merely over the full ranking. We compare the judge's continuous verdict-token score with a supervised correctness probe trained on attention-head activations from the same prompt. Exact identities link fixed-budget error recall to residual risk and show that AUROC averages recall over all budgets. Across three open 7-12B models on GSM8K and MATH, the probe finds 1.3-22.5 additional percentage points of errors at 5% review in all six model-dataset settings (median 7.4 points). At 10% routing, probe-guided allocation improves accuracy over equal-cost random allocation in both GSM8K experiments, but beats judge-guided routing only for the weaker judge. Cross-dataset transfer lowers probe AUROC for every model in both directions relative to target-domain training. Because dataset, weight precision, and generation budget change together, this measures a joint protocol shift rather than a dataset-only effect. Internal signals can improve supervised, in-domain budget allocation, but are not domain-general confidence estimators.

Figures & tables

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. LLM-as-Judge on a Budget

    Feb 17, 2026Aadirupa Saha, Aniket Wagde, Branislav KvetonLLM EvaluationLLM-as-a-Judge

  2. Judge, Retrieve, or Abstain: Uncertainty-Guarded LLM Judging with Provable Risk Guarantees

    Aug 18, 2026Sher Badshah, Ali Emami, Hassan SajjadLLM-as-a-JudgeSelective Prediction

  3. Reasoning Is Not Free: Robust Adaptive Cost-Efficient Routing for LLM-as-a-Judge

    May 11, 2026Wenbo Zhang, Lijinghua Zhang, Liner Xiang +1LLM-as-a-JudgeCost-Aware Inference